Re: Portals vs Links

Mark Waks ([email protected])
Thu, 11 May 95 13:24:07 EDT


Linas writes:
>I guess I can't quite tell if you are making an artistic comment, or a
>technical comment. Certainly today, you can draw a large rectangle
>(doorway), and use WWWAnchor so that when clicked on, you get to the
>other side. I guess that in the back of my mind, I always assumed
>that the place you came from (the geomterical data) would still "be
>there" (i.e. still be in your browser). Thus, you could
>dolly/pan/zoom around with your camera, and lo& behold, the old place
>was still there, back behind the door.

Sorry, but I don't see it. What door? The current WWWAnchor is a
strictly one-way street -- the *is* no door defined on the other
side. I suppose that the browser could *insert* a return door at
the point where you enter. I've got mixed feelings on that: on the
one hand, it might buy you most of the flavor of portals within the
current spec; on the other, it seems like a *serious* cheat, and
abuses the semantics of the scene graphs. (That is, if there's no
door specified at the other end, inserting one into the scene just
seems a bit wrong.)

It could probably be done, but it certainly doesn't match my impression
of what the spec is talking about. (If it *is* what the spec intends,
I'd appreciate some clarification in the spec; I'd always gotten the
impression that the spec uses WWWAnchor to "jump" between worlds, a la
a click in HTML...)

>As to "wide open spaces", I beleive the "correct" way of doing these
>is to WWWInline whatever you want to put into these spaces, and to
>surround the WWWInline with level-of-detail constructs, so that if the
>object is very far away (small), you don't inline anything, you merely
>draw a small-- ah triangle, whatever. By not inlining, you save the
>network traffic.

Could work, but it still requires an extremely "macro" approach to
the problem -- you still have to have a "top-level" scene graph that
describes the thing overall. Seems like a poor way to describe
something as intrinsically distributed as this problem. (In general,
forcing the browser to create large "global spaces" internally strikes
me as inappropriate, since the user is generally only concerned with
relatively "local" space.)

I will admit that it's possible, though, and hadn't occurred to me; I
assume that this is how they're doing the Virtual SoMa...

-- Justin

Random Quote du Jour:

Re: Trusting Your Sources
"Next thing I know you'll be telling me that `Eureka' doesn't translate to,
`This bath is too hot.'"
-- Aindries